RE: filmscanners: pixels, printer dots, etc

["michael shaffer" <rarewolf@roadrunner.nf.net>]

Austin writes ...

shAf writes ...
> > which may beg the question ... "If I want to avoid
> > resampling artifacts, is there an optimum res to send
> > to the printer?"  In some cases the answer is "yes".
>

> What cases do you know of (WRT modern inkjet printers that is)?
>
> From my experience (WRT modern inkjet printers), the answer has
> always been no...

then Austin writes ...

> why do you believe "the software has to resample" when creating
> a dither pattern?

  Last Q 1st ... If the res of your image is (e.g.) 300PPI, and you indicate
"draft", "good" or "best", then there needs to be some translation.  If
there is a printer specific optimum resolution associated with the mode I
would assume the printer software would need to resample.

  I have only tested my printer (Epson 1280), but the reasons why I tested
it came from generalizations of all Epson softwares (Macs and Pro series
printers included).  The test is to create an image from which rasampling
artifacts should be obvious, and send it to the printer with varying PPI
settings forcing the printer to resample (if it indeed does) ... and examine
the print.

  This is all that is necessary for recognizing Epson resampling artifacts:

(1)  Create a 180PPI square image (gray, RGB) with almost vertical and
horizontal solid 1 pixel black on white lines.  The angle of the lines
should allow the line segments to be 5-10 pixels long and composed of 10-20
steps.  You can play with colored lines later ... for now black on white
and/or white on black.

(2)  Configure the printer software for your usual "photographic" settings,
which is "print quality" 720x720 or better, "microweave" enabled, "high
speed" disabled, "finest detail" disabled (sharp edges only), and "smooth
edges" disabled (for upsampling lo-res dig-cam images).

(3)  With these print settings, print the image at these PPI resolutions:
180, 240, 300, 360.  That is, re-size with 'resample' disabled.

  I'll leave it to you to let us know which resolutions show you predicted
printed pixels.  But what I saw was different line segments being different
widths, which are resampling and anti-aliasing artifacts.  The fact that
these artifacts do not show up for specific PPI resolutions would imply
there is an optimum hardware resolution, and the printer software resamples
before printing.

  I didn't believe this either ... printer softwares should be better than
this.  With Photoshop, you can even resample to this "optimum" res before
sending it to the printer and do a better job (but not perfect).  If you do
some resampling experiments with PS, you realize Epson is using something
similar to "bilinear", or at least realize the artifacts are similar.

  Lastly I'll grant you ... this is NOT a "photographic" image, and it is
doubtful you or I would ever be able to point to a resampling artifact in an
actual printed photographic image ... but I think you'll admit after testing
they must be there.

shAf  :o)